Gemma 3 4B IT
Gemma 3 introduces multimodality, supporting vision-language input and text outputs.
google.gemma-3-4b-itReleased 13 Mar 2025 (Mar 2025)Pricing
List prices at Amazon Bedrock, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $0.04 | $0.08 | |
| Batch 50% off | $0.02 | $0.04 | 50% off |
| Flex 50% off | $0.02 | $0.04 | 50% off |
| Priority 75% more | $0.07 | $0.14 | 75% more |
What it costs a month
At Amazon Bedrock list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out$18.00
- RAG over documents2,000 requests a day, 8K in / 0.6K out$22.08
- Coding agent1,000 requests a day, 40K in / 2K out$52.80
- Overnight batch20,000 requests a day, 4K in / 0.4K out, batch$57.60
Where to run it
3 hosts serve Gemma 3 4B IT. The comparison is on blended price, 3 parts input to 1 part output.
| Host | In / out per 1M | Model id | |
|---|---|---|---|
Amazon BedrockCloud google.gemma-3-4b-it | $0.04 / $0.08 Retires 30 Mar 2027 | google.gemma-3-4b-it | Retires 30 Mar 2027 |
DeepInfraInference gemma-3-4b-it | $0.05 / $0.10 +25% vs headline | gemma-3-4b-it | |
NVIDIA NIMInference gemma-3-4b-it | - | gemma-3-4b-it | |
LemonadeLocal Gemma-3-4b-it-GGUF | - | Gemma-3-4b-it-GGUF |
Show 1 more listings: regions, snapshots and variantsHide regional listings
| Amazon Bedrock | |||||||
| Mantle google.gemma-3-4b-it | $0.04 / $0.08 Retires 30 Mar 2027 | google.gemma-3-4b-it | Retires 30 Mar 2027 | ||||
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- TextImages
- Produces
- Text
- Open weights
- License: Gemma LicenseRuntimes: Ollama, llama.cpp, vLLM
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
No host sells it for less than Amazon Bedrock.
Smaller Google models
No cheaper Google model keeps its key capabilities.
Other frontier labs
No cheaper match from another frontier lab.
Open-weight picks
- 3% less
- 15% less
- 20% less
- 19% less
Call it
Using Amazon Bedrock's API.
from litellm import completion
response = completion(
model="bedrock/google.gemma-3-4b-it",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does Gemma 3 4B IT cost?
On Amazon Bedrock, Gemma 3 4B IT costs $0.04 per million input tokens and $0.08 per million output tokens.
Which APIs offer Gemma 3 4B IT?
Gemma 3 4B IT runs on 3 APIs: Amazon Bedrock, DeepInfra and NVIDIA NIM.
What is Gemma 3 4B IT's context window?
128K tokens, with up to 8K tokens in one reply.
Is Gemma 3 4B IT being retired?
Amazon Bedrock retires it on 30 Mar 2027. 2 hosts keep serving it.
What are cheaper alternatives to Gemma 3 4B IT?
Models that cost less and keep most of what it can do: Qwen3.7 Flash Variant1 on EmpirioLabs AI (3% less). Savings are on a mix of 3 input tokens to 1 output token.
When was Gemma 3 4B IT released?
Google released it on 13 Mar 2025.
What is Gemma 3 4B IT's API model id?
Use google.gemma-3-4b-it with Amazon Bedrock. Other hosts use their own ids: gemma-3-4b-it on DeepInfra, gemma-3-4b-it on NVIDIA NIM.
Sources
Raw catalog entry for google.gemma-3-4b-it on Amazon Bedrock
{
"deprecation_date": "2027-03-30",
"deprecation_source": "https://docs.aws.amazon.com/bedrock/latest/userguide/model-lifecycle-legacy.html",
"input_cost_per_token": 4e-8,
"input_cost_per_token_batches": 2e-8,
"input_cost_per_token_flex": 2e-8,
"input_cost_per_token_priority": 7e-8,
"litellm_provider": "bedrock_converse",
"max_input_tokens": 128000,
"max_output_tokens": 8000,
"mode": "chat",
"output_cost_per_token": 8e-8,
"output_cost_per_token_batches": 4e-8,
"output_cost_per_token_flex": 4e-8,
"output_cost_per_token_priority": 1.4e-7,
"source": "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/current/us-east-1/index.json",
"supports_audio_input": false,
"supports_function_calling": true,
"supports_system_messages": true,
"supports_vision": true
}